GFRP层板缺陷线性调频红外热波成像检测概率和特征图像融合算法研究

Glass fiber reinforced polymer (GFRP) is one of the best modern ship materials for the pursuit of higher payload and stealth performance. However, the complex and harsh application environments require non-destructive testing technologies for GFRP component defect and damage having higher detectability and reliability. The purpose of this project mainly focuses on the reliability evaluation and characteristic image fusion algorithm for GFRP laminate defect inspection combining linear frequency modulated infrared thermal wave imaging (LFMITWI), probability analysis of defect detection and image fusion technique. The thermal wave response characteristics and noise jamming mechanism of GFRP laminate defect testing using LFMITWI will be first studied. The suppressing method of the infrared image noise and the novel fast extraction algorithms will be investigated subsequently for constructing the characteristic image of the linear frequency modulated thermal wave signal (LFMTWS) in time-frequency domain. The probability density distribution model of defect detection and the influence of detection noise level, testing parameters and characteristic extraction algorithms on the defect detection probability level will be determined with statistical analysis and testing experiments. The influence of fusion algorithm on the detection performance with a usability evaluation of defect detection probability level and the optimal design method of the fusion operator will be deeply studied based on the modern intelligent optimization algorithms. This project will provide a new approach for the comprehensive improvement of the detection performance and reliability of non-destructive testing for GFRP laminate defect using LFMITWI, and also provide an important reliability judgment basis for the future engineering applications.

玻璃纤维增强复合材料(GFRP)是现代舰船追求更高有效载荷及更强隐身性能的最佳材料之一，复杂恶劣的服役环境对GFRP部件内部缺陷无损检测的探测性能及可靠性提出了更高要求。项目旨在采用线性调频红外热波成像、缺陷检测概率分析与图像融合技术相结合的方法，开展GFRP层板线性调频红外热波成像无损检测可靠性评价和特征图像增强融合算法研究。研究线性调频红外热波成像检测GFRP层板缺陷的热波响应特性及噪声干扰机制；提出线性调频热波信号时频特征提取算法；建立缺陷检测的概率密度分布模型，探讨检测噪声水平、检测参数和特征提取算法对缺陷检测概率水平的影响；以缺陷检测概率水平为融合效果评价指标，研究特征图像融合算法对缺陷探测性能的影响，运用现代智能优化算法实现对融合算子参数的快速寻优。研究工作可为GFRP层板缺陷线性调频红外热波成像综合探测性能的提升提供新方法，为该技术的工程应用提供无损检测可靠性评判依据。

玻璃纤维增强复合材料(Glass Fiber Reinforced Polymer，GFRP)层板在成型和使用过程易产生空隙、分层和脱粘等缺陷，严重影响GFRP制件的结构安全性和使用寿命，在生产过程中迫切需要简捷、高效的无损检测方法实现对GFRP层板缺陷的可靠检测。本项目将线性调频红外热波成像、缺陷检测概率分析与图像融合技术相结合，开展了GFRP层板线性调频红外热波成像无损检测可靠性评价和特征图像增强融合算法研究。研究了线性调频热流在GFRP层板中热波扩散特征，仿真分析了线性调频红外热波成像检测GFRP层板缺陷的热波响应及噪声干扰特性；提出了基于Chirp相关的时间切片特征信息提取算法，分别搭建了制冷和非制冷型线性调频红外热波成像检测系统，试验研究表明基于Chirp相关的时间切片特征提取算法可有效提高检测图像的信噪比；建立了缺陷检测的概率密度分布模型，基于POD（Probability of Detection）分析方法研究了常用的特征提取算法和噪声干扰对检测概率水平的影响，研究表明基于Chirp相关的时间切片特征提取算法具有最佳的缺陷检出率和可靠性水平；基于小波变换研究了特征图像增强和融合算法对探测性能的影响，采用模拟退火算法实现了对基于小波变换的图像增强和融合算子参数的快速寻优，研究表明基于小波变换的图像增强和融合算法可有效抑制噪声干扰，显著提升线性调频红外热波成像检测技术的探测性能。研究成果为GFRP层板缺陷线性调频红外热波成像检测技术的可靠性评价和综合探测性能的提升提供了新方法，为该技术的工程应用提供了无损检测可靠性评判依据。